It is generally considered that the human speech signal consists of discrete units. However, the acoustic nature of the individual speech sound varies considerably depending on its context in continuous speech signal. It is also seen in many situations that the acoustic manifestation of each unit is overlapping in the time domain. Some of the typical examples of those phenomena are reviewed and their characteristics are briefly discussed with reference to the movements of the articulatory organs.
Smoothing of the movement of the articulatory organs due to the inertia and friction is apparently one of the important factors that affect the acoustic nature of the speech sound. Besides this, there are other kinds of effects, such as coarticulation effect, which appear to be more specific to the speech production process. It is considered that the direct observations of articulatory conditions would be essential for the better understanding of the structure of the human speech signal. Recent experimental techniques for the observation of the articulation and phonation are presented.

抄録全体を表示

Many kinds of muscle models have been reported, but none of them goes beyond explanation of specified phenomena in muscular contraction. In order to clarify the contraction mechanism quantitatively, development of a model capable of simulating contractile phenomena under various physiological conditions has been desired.
A new viscoelastic model of skeletal muscle, which consists of the elastic, viscous and the contractile components is proposed on the structural basis of the muscle cell and the sliding-filaments theory of muscular contraction. Some physiological experiments were made on frog's semitendinosus muscle in Ringer solution to determine the characteristics of each component of the model from input-output relation by the simulation technique.
The characteristics of the passive elements, e. g., the viscous and elastic components of resting muscle, were determined through simulations of the experiments on quick loading and constant velocity stretching. The non-linear characteristics of the series elastic component of the contracting muscle were estimated as a function of extension by analyses of the isotonic and isometric quick release techniques. The results were proved reasonable through the analog computer simulations of Buchthal's (1944) vibration experiment and Hill's (1949) quick stretch experiment.

抄録全体を表示

The authors have already proposed a new viscoelastic model which contains the viscous, elastic and contractile components. In this paper, some physiological experiments made on frog's semitendinosus muscle in Ringer solution in order to determine the dynamic constants of the viscous and contractile components of this model in contracting muscle are presented. Using an analog computer, these experimental results were simulated on the basis of the model.
The characteristics of the viscous component are estimated as a function of velocity and force of contraction (the internal contractile force generated in the contractile component) through analysis of the experiments of isotonic contraction and constant velocity stretching. When the contractile component was stretched at a constant velocity, the viscous force increased linearly with the force of contraction. The viscous characteristics obtained from the isotonic contraction agreed exactly with the simulated results of the isotonic quick release experiment.
It was clarified by the calculation based on the model that the active state determined by the quick release technique represented the internal contractile force generated in the contractile component. The quick release experiments were simulated with good agreement by the model.
Through these simulations, the validity of the viscoelastic model has been proved.After all, the time courses of the internal contractile force during isometric twitches were estimated in the frog's semitendinosus muscle using the characteristics of the viscous and elastic components of the model.

抄録全体を表示

In this paper the results of computer simulation of the automatic interpretation system for chest X-ray films are presented.
First the outline of the system is stated. The system consists of four subsystems ; subsystem I for recognition of rib images, subsystem II for elimination of rib images, subsystem III for equalization of average density level in the horizontal direction, subsystem IV for recognition of abnormal shadows. A program system capable of simulating the whole performance of the interpretation system has been developed and used for the simulation described here.
Next, the results of recognition experiment of abnormal shadows are shown. Part of a chest X-ray film including tuberculous lesion were used as an input pattern for the system. All the rib images were recognized satisfactorily. In the recognition of abnormal shadows, however, a few errors occurred.
Lastly, problems to be solved in the future are discussed in detail.

抄録全体を表示